---
title: "How to interpret the HTML report generated by `cellCellReport` function"
author:
- name: Koki Tsuyuzaki
affiliation: Laboratory for Bioinformatics Research,
RIKEN Center for Biosystems Dynamics Research
- name: Manabu Ishii
affiliation: Laboratory for Bioinformatics Research,
RIKEN Center for Biosystems Dynamics Research
- name: Itoshi Nikaido
affiliation: Laboratory for Bioinformatics Research,
RIKEN Center for Biosystems Dynamics Research
email: k.t.the-answer@hotmail.co.jp
package: scTensor
output:
BiocStyle::html_document
vignette: |
%\VignetteIndexEntry{scTensor: 2. Interpretation of HTML report}
%\VignetteEngine{knitr::rmarkdown}
%\VignetteEncoding{UTF-8}
---
# Introduction
Here, we explain the way to interpret of HTML report generated by `cellCellReport`.
If `cellCellDecomp` is properly finished, we can perform `cellCellReport`
function to output the HTML report.
The results can be confirmed by typing `example(cellCellReport)`.
The report will be generated in the temporary directory (it costs 5 to 10 minutes).
The output directory contains some files and directories as follows.
- **index.{Rmd,html}** : The main HTML report
- **reanalysis.RData** : The R binary file for using in the HTML report and reanalysis of scTensor
- **Workflow_2.png** : The figure for the section 1. About scTensor Algorithm in the HTML report
- **figures** : The directory containing some figures for the HTML report
- **ligand.{Rmd,html}** : The HTML report for the section 6. Gene-wise Hypergraph in the HTML report
- **ligand_all.{Rmd,html}** : The HTML report for the section 6. Gene-wise Hypergraph in the HTML report
- **receptor.{Rmd,html}** : The HTML report for the section 6. Gene-wise Hypergraph in the HTML report
- **receptor_all.{Rmd,html}** : The HTML report for the section 6. Gene-wise Hypergraph in the HTML report
- **pattern_X_Y.{Rmd,html}** : The HTML report for For the section 7. (Ligand-Cell, Receptor-Cell, ) -related L-R Pairs in the HTML report
Here, look at the index.html.
# Interpretation of "1. About scTensor Algorithm"
In the HTML report, the 1st item describes the overview of `r Biocpkg("scTensor")` and other CCI-related packages.
# Interpretation of "2. Global statistics and plots"
The 2nd item describes all the R objects saved in **reanalysis.RData**,
which contains the result of `r Biocpkg("scTensor")`.
This file is saved in the output directory (**out.dir**) specified in `cellCellReport`,
and the user also can re-analyze the result of `r Biocpkg("scTensor")`.
Using `r CRANpkg("plotly")` package, `cellCellReport` generates some interactive plots.
For example, in item 2.1, the number of cells in each cell type can be confirmed when the cursor moved on the box.
In item 2.2, the number of expressed genes in each cell type (Non-zero genes) can be confirmed when the cursor moved on the box.
In item 2.3, the two-dimensional plot user specified can be confirmed.
In item 2.4, the distribution of core tensor values and the value of each (Ligand-Cell-type, Receptor-Cell-type, LR-pair) pattern can be confirmed.
The red bars mean that these values are selected by the threshold (**thr** parameters) in `cellCellReport`.
Note that the thr can be specified from 0 to 100,
the large thr value will generate too many HTML files (cf. 8. (Ligand-Cell, Receptor-Cell, LR-pair) Patterns)
and takes a long time.
The 3-order CCI-tensor consisting of Cell_L $\times$ Cell_R $\times$ LR-pair (LR) are decomposed by `r CRANpkg("nnTensor")`,
in which the tensor is iteratively matricised to mode-1 (Ligand-Cell direction) and mode-2 (Receptor-Cell direction).
In each direction, NMF is performed and the strength of each directional pattern is summarized in the bar plots.
For example, in item 2.5, the distribution of mode-1 matricised tensor can be confirmed.
Likewise, in item 2.6, the distribution of mode-2 matricised tensor can be confirmed,
# Interpretation of "3. Ligand-Cell Patterns"
In the 3rd item, using the heatmap of `r CRANpkg("plotly")`, the user can interactively confirm the detail of Ligand-Cell Patterns extracted by `r CRANpkg("nnTensor")`.
# Interpretation of "4. Receptor-Cell Patterns"
Likewise, in the 4th item, the user can interactively confirm the detail of Receptor-Cell Patterns.
# Interpretation of "5. CCI-wise Hypergraph"
In the 6th item describes, the strength between Ligand-Cell Patterns and Receptor-Cell Patterns (CCI-strength),
by the summation of the core tensor with the mode-3 direction,
a matrix consisting of the number of Ligand-Cell Patterns $\times$ the number of Receptor-Cell Patterns.
# Interpretation of "6. Gene-wise Hypergraph"
In the 7th item, the relationship between LR-pairs, which coexpressed in any LR-pair pattern at least one time.
Ligand genes are described as red nodes,
receptor genes are described as blue nodes,
and corresponding LR-pair patterns are described as the color of edges.
Using `r CRANpkg("visNetwork")` package,
these interactions can be interactively visualized.
Under the gene-wise hypergraph, four hyperlinks are embedded.
In the 1st link, the details of the gene-wise hypergraph can be confirmed as a corresponding table in a ligand gene-centric manner.
This page can work as a reverse lookup search by "Ctrl + F";
by typing the gene name of ligand that the user is interested in,
the partner receptors, which are coexpressed in some LR-pair patterns,
also can be found.
In the 2nd link, the user can find all the partner receptors,
even if the partner receptors are not coexpressed in any LR-pair pattern,
and if they are not included in the data matrix.
Likewise, the receptor gene-centric reverse search page is embedded in the 3rd link,
and, in the 4th link, all the partner ligand genes are included.
# Interpretation of "7. (Ligand-Cell, Receptor-Cell, LR-pair) Patterns"
In the 8th item, the details of (Ligand-Cell, Receptor-Cell, LR-pair) Patterns are ordered by the size of the core tensor, and the link of each pattern is embedded.
(Note that the number of links is dependent on the **thr** parameter of `cellCellReport`.)
For example, the 1st link describes the details of (3,2,) Pattern, which means the relationship of *1*st pattern of Ligand-Cell patterns, *1*st pattern of Receptor-Cell patterns, and *5*th pattern of LR-pair patterns.
In this pattern, only one LR-pair is coexpressed (INSL3 and GNG11).
The hyperlinks to many databases and PubMed are also embedded.
The degree of the size of the LR-pair in the LR-pair pattern is quantified as P-value and Q-value.
Under the LR-pair list,
the results of many enrichment analysis are also embedded such as Gene Ontology (BP/MF/CC), Reactome, MeSH...etc.
User can confirm the detail of the result of `r Biocpkg("scTensor")`, and perform the biological interpretation.
# Session information {.unnumbered}
```{r sessionInfo, echo=FALSE}
sessionInfo()
```